Machine learning in trading: theory, models, practice and algo-trading - page 2216
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
What a surprise, now I tested the bot published by Maxim Vladimirovich.
In the first screenshot the quotes from Dukascopi, the rising balance line after 3200 is just the period November 2019 - October 2020, 1 hour timeframe, spread - 2 points (0.0002)
The test at another broker in Metatrader gave not so remarkable result, it seems that the strategy should be optimized so that it works for all without exception.
Yes, the distributions usually show everything. You can just do these for the traits\targets without the booster and see right away.
So the idea is just to evaluate the model, and the model actually unravels the tangled targets, and we can evaluate its success in doing so, rather than just seeing how tangled everything is.
I'm thinking of trying the cascade learning method (I made up the term myself - maybe there's something different). The graphs show that there are areas where training is successful - this area to leave, and what goes beyond this area to train again, preliminarily removing from the sample examples caught in the distribution of the left area. I already tried to do it manually and the effect was good, I think to automate it, but the second day I still can't. I'm afraid that the effect was accidental and I don't want to get upset. What is your opinion on this? I think it is easy to do in python.
So the idea is just to evaluate the model, and the model actually unravels tangled targets, and we can evaluate its success in doing so, rather than just seeing how tangled everything is.
I'm thinking of trying the cascade learning method (I made up the term myself - maybe there's something different). The graphs show that there are areas where training is successful - this area to leave, and what goes beyond this area to train again, preliminarily removing from the sample examples caught in the distribution of the left area. I already tried to do it manually - the effect was good, and I think to automate it, but the second day I still can't - I'm afraid that the effect was accidental - I do not want to get upset. What is your opinion on this? I think that in python it is easy to do.
Well, it's all a question of semi-controlled learning. While I'm reading
So the idea is just to evaluate the model, and the model actually unravels tangled targets, and we can evaluate its success in doing so, rather than just seeing how tangled everything is.
I'm thinking of trying the cascade learning method (I made up the term myself - maybe there's something different). The graphs show that there are areas where training is successful - this area to leave, and what goes beyond this area to train again, preliminarily removing from the sample examples caught in the distribution of the left area. I already tried to do it manually - the effect was good, and I think to automate it, but the second day I still can't - I'm afraid that the effect was accidental - I do not want to get upset. What is your opinion on this? I think that in Python it is easy to do.
If you can split it into homogeneous areas automatically, just like with the hands plus minus, it will work.
What a surprise, now I tested the bot published by Maxim Vladimirovich.
In the first screenshot the quotes from Dukascopi, the rising balance line after 3200 is just the period November 2019 - October 2020, 1 hour timeframe, spread - 2 points (0.0002)
If I look at the results of MetaTrader I will see that the strategy has to be optimized so that it works for everyone without exception.
It is not the best variant. If you look at the settings you may get even better.
too lazy to write new metrics yet... and it will definitely not be max profit then.....
you can't do that, it's just not there
You can't do that, it's just not there.
I can do it, but you have to learn how to use google.
like catboost custom loss functionFar from the best option. Pick up the settings, you can get these and even better
With these calculations (changing only two parameters look_back and ma_periods) what is the load on the processor approximately?
with these calculations (changing only two parameters look_back and ma_periods) approximately what is the load on the processor?
do not know, do not notice
I can do it, but you have to learn how to use Google.
e.g. catboost custom loss functionYou don't understand, try it, you have a function that calculates the balance...
it will take you less than a minute to figure it out